Many of us are still looking at politics through an obsolete, one-dimensional lens that projects the political representatives and their votes on a left-right spectrum. It is of course much more complex, but how can we assess that complexity, and objectively describe how our representatives and parties are positioned with respect to each other?
This work is an attempt at seeing how our MEPs shape the EU political landscape through their votes, regardless of their party affiliation. This is done by applying data mining techniques to voting data published by the European Parliament and http://parltrack.euwiki.org.
The code for this project is hosted as a set of python notebooks on GitHub. In particular :
All the original data and the derived data produced in the scope of this work are licensed under the ODbL v1.0 license.
The MEPs votes data have been processed using Principal Components Analysis, allowing us to reduce 10000+ votes into a few axis that best represent MEPs behaviour. By plotting MEPs along those axis, we can see how they relate to each other, and how they differ, forming the EU political space. It also pictures how countries or parties are positioned within that space.
MEP dots are sized after their attendance over the past 5 years (attendance percentage is displayed in MEP info popups). Lower attendance is most often due to MEP not having been in office for the full 5 years term.
Each political group seems to have a trail of smaller dots drawn towards the center of the chart. This is because MEPs with smaller dots have less actual votes, and missing votes are assimilated to abstention in the scope of the data mining algorithm. This results in MEPs with less votes being drawn to a neutral position.
Axis 1 appears to be the traditional left to right political axis (albeit reversed here), with the PPE at the left, and the radical left to the right of the graph.
Data mining results give us the votes that are most decisive in positioning an MEP along axis 1.
The top 10 of these votes are:
(the vote column links to the visual vote summary, the title column links to the EU Parliament portal with all the related documents)
Looking into those votes, we can see that they are about the TTIP, rail public transports, Trade in Services Agreement, the Right2Water initiative.
Likewise, we can look at the top votes defining Axis 2 to understand what it means :
These votes cover refugees in the labour market, gender mainstreaming, implementation of a European microfinance facility, human rights and migration ...
It is worth noticing that the mere title of the vote may not reflect the actual stake of the vote. In the case of the microfinance facility (vote 61995), the vote is actually about a specific paragraph (§ 28/2), and a closer look at the parliament minutes reveals that it was all about keeping or removing the term "refugees" from the sentence "Calls on the Commission to view refugees and asylum seekers as a target group".
Axis 3 is interesting in that it shows a common behaviour shared by S&D and ECR (and to some extent EFDD and ENF). Looking at the decisive votes for that axis :
we see topics such as fight against fraud (where S&D proposed 4 amendments backed by ECR, see Am 11->14). It is however harder to figure out the common denominator right away.
Likewise, the same investigative work can be carried out for other axes, but requires a close look at every amendment at stake to understand the common denominator to the votes. This is beyond this project, but would certainly be worth the effort.